Lakehouse for Data Engineers
A lakehouse stores data as open table formats such as Apache Iceberg, Delta Lake or Apache Hudi on object storage, with a catalog and query engines such as Trino or Spark on top. This unit explains how those layers provide warehouse features like transactions and schema evolution on cheap storage.
Stage: Lakehouse and Storage · Part of the Data Engineer roadmap · Last reviewed 19 November 2026
What to learn in Lakehouse
- Explain how Iceberg uses metadata files and snapshots to track table state
- Compare Iceberg, Delta Lake and Hudi on features and ecosystem fit
- Run a time-travel query against an earlier snapshot
- Describe the role of a catalog such as Hive Metastore, Glue or a REST catalog
- Choose a query engine for interactive SQL versus heavy batch jobs
Lakehouse interview questions
Why use an open table format instead of plain Parquet files in a bucket?
Plain files lack atomic commits, so readers can see partial writes, and schema changes or deletes are painful. Table formats add a metadata layer giving transactions, snapshots, schema evolution and file-level pruning while keeping the data in open files.
What does a catalog do in a lakehouse?
The catalog maps table names to the current metadata location and handles atomic updates to that pointer. Engines such as Spark, Trino and Flink share it, so they all see the same table version and permissions.
Where Lakehouse fits in the Data Engineer roadmap
Lakehouse belongs to the Lakehouse and Storage stage (intermediate level), which takes about 3 hours. Open table formats.
Back to the full Data Engineer roadmap